2. The Basic Order Typology*/4

Linguists are, in general, familiar with the notion that certain languages tend consistently to put modifying or limiting elements before those modified or limited, while others just as consistently do the opposite. Turkish, an example of the former type, puts adjectives before the nouns they modify, places the object of the verb before the verb, the dependent genitive before the governing noun, adverbs before adjectives which they modify, etc. Such languages, moreover, tend to have postpositions for concepts expressed by prepositions in English. A language of the opposite type is Thai, in which adjectives follow the noun, the object follows the verb, the genitive follows the governing noun, and there are prepositions. The majority of languages, as for example English, are not as well marked in this respect. In English, as in Thai, there are prepositions, and the noun object follows the verb. On the other hand, English resembles Turkish in that the adjective precedes the noun. Moreover, in the genitive construction both orders exist: 'John's house' end 'the house of John'.

More detailed consideration of these and other phenomena of order soon reveals that some factors are closely related to each other while others are relatively independent. For reasons which will appear in the course of the exposition, it is convenient to set up a typology involving certain basic factors of word order. This typology will be referred to as the basic order typology. Three sets of criteria will be employed. The first of these is the existence of prepositions as against postpositions. These will be symbolized as Pr and Po, respectively. The second will be the relative order of subject, verb, and object in declarative sentences with nominal subject and object. The vast majority of languages have several variant orders but a single dominant one. Logically, there are six possible orders: SVO, SOV, VSO, VOS, OSV, and OVS. Of these six, however, only three normally occur as dominant orders. The three which do not occur at all, or at least are excessively rare, are VOS, OSV, and OVS. These all have in common that the object precedes the subject. This gives us our first universal:

Universal 1. In declarative sentences with nominal subject and object, the dominant order is almost always one in which the subject precedes the object.5
This leaves us with three common types: VSO, SVO, and SOV. These will be symbolized as I, II, and III, respectively, reflecting the relative position of the verb.

The third basis of classification will be the position of qualifying adjectives (i.e., those designating qualities) in relation to the noun. As will be seen later, the position of demonstratives, articles, numerals, and, quantifiers (e.g., 'some', 'all') frequently differs from that of qualifying adjectives. Here again there is sometimes variation, but the vast majority of languages have a dominant order. Dominant order with adjective preceding noun will be symbolized by A and dominant order with noun preceding adjective by N. We thus arrive at a typology involving 2 x 3 x 2, that is, twelve logical possibilities. The 30 languages of the sample are distributed among these twelve classes as shown in Table 1.6

TABLE 1
 
  I II III
Po-A 0 1 6
Po-N 0 2 5
Pr-A 0 4 0
Pr-N 6 6 0
The table has been arranged so that the 'extreme' types Po-A and Pr-N are in the first and fourth row, respectively. It is evident that with respect to these extremes, I and III are polar types, the former being strongly correlated with Pr-N and the latter with Po-A. Type II is more strongly correlated with Pr-N than with Po-A. It is also clear that adjective position is less closely related to types I, II, and III than is the Pr/Po contrast. The table is, I believe, a fair representation of the relative frequency of these alternatives on a world-wide basis. Type II is the most frequent; type III almost as common; type I is a definite minority. This means that the nominal subject regularly precedes the verb in a large majority of the world's languages.

Turning for a moment to genitive order, we note that this characteristic might fittingly have been utilized for typological purposes. The reason for not employing it is its extremely high correlation with Pr/Po, a fact generally known to linguists. It would thus virtually have duplicated the latter criterion. It was not chosen because Pr/Po on the whole is slightly more highly correlated with other phenomena. Of the present sample of 30 languages, 14 have postpositions, and in every one of these the genitive order is genitive followed by governing noun. Of the 14 prepositional languages, 13 have the genitive following the governing noun. The only exception is Norwegian, in which the genitive precedes. Thus, 29 of the 30 cases conform to the rule. If anything, 1/30 is an overestimation of the proportion of exceptions on a world-wide basis. We therefore have the following universal:

Universal 2. In languages with prepositions, the genitive almost always follows the governing noun, while in languages with postpositions it almost always precedes.
Turning once more to the data of Table 1, we find striking evidence of lawful relationships among the variables in that of the 12 possibilities 5, or almost half, are not exemplified in the sample. All of these types are either rare or nonexistent.7 For type I, we see that all 6 languages of the sample are Pr/N. This holds with extremely few exceptions on a world-wide basis. There are, however, a few valid examples of I/Pr/A, the mirror image, so to speak, of the fairly frequent III/Po/N. On the other hand, there are, as far as I know, no examples of either I/Po/A or I/Po/N. Hence we may formulate the following universal: Languages of type III are, as has been seen, the polar opposites of type I. Just as there are no postpositional languages in type I, we expect that there will be no prepositional languages in type III. This is overwhelmingly true, but I am aware of several exceptions.8 Since, as has been seen, genitive position correlates highly with Pr/Po, we will expect that languages of type III normally have GN order. To this there are some few exceptions. However, whenever genitive order deviates, so does adjective order, whereas the corresponding statement does not hold for Pr/Po.9 We therefore have the following universals:
Universal 4. With overwhelmingly greater than chance frequency, languages with normal SOV order are postpositional.

Universal 5. If a language has dominant SOV order and the genitive follows the governing noun, then the adjective likewise follows the noun.

An important difference may be noted between languages of types I and III. In regard to verb-modifying adverbs and phrases as well as sentence adverbs, languages of type I show no reluctance in placing them before the verb so that the verb does not necessarily begin the sentence. Further, all VSO languages apparently have alternative basic orders among which SVO always figures. On the other hand, in a substantial proportion, possibly a majority, of type III languages, the verb follows all of its modifiers, and if any other basic order is allowed, it is OSV. Thus the verb, except possibly for a few sentence modifiers (e.g., interrogative particles), is always at the end in verbal sentences. It is not logically required, of course, that languages all of whose basic orders involve the verb in the third position should also require all verb modifiers to precede the verb, but this seems to hold empirically. Thus, languages in which the verb is always at the end may be called the "rigid" subtype of III. In the present sample, Burushaski, Kannada, Japanese, Turkish, Hindi, and Burmese belong to this group, while Nubian, Quechua, Basque, Loritja, and Chibcha do not.10 These considerations permit us to state the following as universals:
Universal 6. All languages with dominant VSO order have SVO as an alternative or as the only alternative basic order.

Universal 7. If in a language with dominant SOV order, there is no alternative basic order, or only OSV as the alternative, then all adverbial modifiers of the verb likewise precede the verb. (This is the rigid subtype of III.)

*Joseph H. Greenberg, "Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements",  In: Joseph H. Greenberg (ed.). Universals of Language. London: MIT Press, pp. 76-80.